Mechanical Properteies Of Micro-structure And Analysis Of Reliability

In Mems devices, many components has used the membranous material, the Mems component analysis’s key is analyzing its mechanical properties. Because of receiving the residual stress, elasticity coefficient’s error, Thin film crystal lattice’s variety, Poisson’s ratio, Microstructure’s weary and destruction toughness’s influence and so on, The actual measuring data and the theoretical formula always have some errors.First,this article adopted the actual survey’s data to validate the relationship of the force and deflection under the analysis of the Matlab environment. Secondly, through the BP network and RBF network to predict a set of data, Getting that the forecast of BP network and RBF network between the relationship of deflection and force is feasible, but the prediction rate of the RBF network is faster and prediction results is better. Again, using BP network to make function approximation to parts of the nonlinear, after a lot of experimental training of sets of data, getting the relationship of deflection and force:T=purelin(W2*tansig(W1*P+b1)+b2), Here,P、T respectively is force、deflection vector; W1,W2,b1,b2 respectively is the right value of operation.The approaching function Of the BP network is the better application.